Está en la página 1de 11

Study on Low Reynolds Number Airfoil Design for Micro Air Vehicle (MAV) with Adaptive Wing

Ranggi S. Ramadhan
MSc Aerodynamics and Aerostructures
Department of Mechanical Engineering, University of Sheffield
Abstract
In order to perform efficiently in various working condition, morphing or adaptive wing is now being researched
for Micro aerial vehicle (MAV) application. The purpose of this study is to investigate and obtain the optimum
wing design for MAV at two different flight condition: cruise and loiter. The design includes analysis and
simulation using XFOIL software, and focused on airfoil selection for low Reynolds number application.
Laminar separation bubble and transition point was found and its effect to airfoil performance is discussed.
The result shows that NACA 1203 and 6406 are the optimum airfoil profile for cruise and loiter condition.
simplify the design work, it still gives a good basic
understanding in wing design and the effect of low
Reynolds number regime.

I. Introduction
Micro aerial vehicles or MAVs, is often described as
small, lightweight aircraft that is controlled remotely
by operator. It is built for specific mission such as
surveillance or intelligence for both civil and
military purposes. Propelled mostly by electric
motor, MAVs are expected to operate in wide variety
of speed and working environment, therefore
efficient aerodynamic performance is expected [1]
[2] [3]. In order to meet the expected aerodynamic
performance at different operating condition,
morphing or adaptive wing design is proposed for
future MAV development. Morphing or adaptive
wing described as a wing which has capability to
change its shape during flight, through some
actuators. Some of the method like twist morphing
(TM) was already investigated [4].
Meanwhile, due to its small dimension and low
flying speed, MAVs operate in the low Reynolds
regime compared to the larger, manned aircraft. This
somehow bring undesirable characteristics such as
low lift to drag ratio caused by laminar separation
bubbles (LSB) and transition. Laminar separation
bubbles is caused by the inability of the flow to make
transition to turbulent flow on the surface of airfoil
and instead separates before transition. It
characterized by laminar separation, laminarturbulent transition and turbulent reattachment. This
separation bubbles become one of the main source
of high drag on low Reynolds number airfoil [5] [6]
[7].
This report presents a study in adaptive MAV wing
design for specific operating condition. The study
includes analytical and numerical method in
deciding the optimum wing platform and profile that
satisfy design requirement.
The numerical method is conducted using XFOIL, a
design and analysis software for subsonic isolated
airfoils [8]. The design work includes selection of
wing aspect ratio and optimum 4-digit NACA
profile for cruise and loiter condition. Although
assumptions and idealizations are made to limit and

II. Methods
II.1. Problem definition
The purpose of this study is to find the optimum
wing design for MAV at cruise and loiter condition.
Cruise defined as a flight when the MAV fly in its
normal speed, steady with no acceleration and with
no change in altitude. While loiter is also defined as
a constant speed level altitude flight but with lower
velocity. The upcoming wind is assumed to always
parallel in the direction of the flight, and the wind
speed relative to the MAV speed is always zero.
Therefore, in both cruise and loiter condition, lift is
always the same with aircraft weight and thrust is
always the same with drag, as shown in equation (1a)
and (1b)

induce drag which is desirable, but it also reduce


Reynolds number and lift to drag ratio. Since the lift
required by the design specification is constant,
lower lift to drag ratio would mean an increase in
drag.

Table 1. Design Specification

=
=

(2)

In a three-dimensional wing design, an additional


component of induced drag is considered. Induced
drag is an additional component of drag (beside
pressure and friction drag) that arise as a
consequence of a downward velocity component
so called downwash reducing the effective angle
of attack. This occurs due to wing-tip vortices
phenomenon, one that can only be observed on finite
wing [9]. Its relation to aspect ratio described by
equation (3).

(1a)
(1b)

Meanwhile Table 1 shows the design specification


of the MAV. It is assumed to fly in sea level
condition with given density and viscosity. It has
cruise speed of 15 m/s and loiter speed of 8 m/s. The
weight is 5 N and has wing area of 0.13 m2. The wing
shape is determined to be rectangular with span
efficiency of 0.9. The MAV is treated as a flying
wing, therefore the effect of other aircraft
component to aerodynamics performance is ignored.
Also, it is assumed that the airfoil profile of the wing,
includes its thickness, maximum camber and camber
position can change for not only cruise or loiter
condition, but also for take-off, climb and landing
condition. However, this study would primarily
discus about the wing design in cruise and loiter
condition. The optimum design would aim the
highest L/D value possible, with minimum total drag
and a consideration in a structural issue.
II.2. XFOIL
XFOIL is an interactive analysis and design software
written by Mark Drela from Massachusetts Institute
of Technology. The first 1.0 version was written in
1986 and has been upgraded and modified for
specific application ever since. For inviscid
formulation, XFOIL use linear-vorticity panel
method. And for the trailing edge base thickness is
modelled with a source panel. While for viscous
formulation, the velocity at each point on the airfoil
surface and wake is obtained from vortex panel
solution with Karman-Tsein correction [8]. This
software is used in this study to obtain optimum
wing profile and analysing the boundary layer
phenomenon over tested airfoils.
II.3. Wing platform optimization
In wing design, the term Aspect Ratio (AR) is
defined by equation (2), which is the square of wing
span divided by wing area. The aspect ratio affects
aerodynamic performance in two opposite way:
increasing value of aspect ratio would decrease

, =

(3)

Meanwhile, aspect ratio affects Reynolds number by


changing the chord length. With constant wing area
S, a change in wing span would automatically
change the length of chord. And since chord is used
as the effective length to calculate Reynolds number
of flow over airfoil shown by equation (4), the
change of its value would change the Reynolds
number.
=

(4)

The other consideration on deciding aspect ratio for


aircraft wing is structural issue. The larger the aspect
ratio thus larger wing span, the more the bending
load experienced at the wing root. Therefore, all the
three parameters namely induce drag, lift to drag
ratio and structural issue become the consideration
of deciding wing platform.
To do the optimization, NACA 2412 and 2812 is
tested on XFOIL using a range of Reynolds number.
The Reynolds number itself is obtained by varying
the aspect ratio. From the test, the total drag of each
aspect ratio variation could be obtained. By also
including the structural issue, the aspect ratio then
can be decided.
II.4. Panel independence study
The underlying principal of XFOIL is panel method,
where a finite number of panel is distributed to
represent the shape of airfoil. This would produce an

Figure 1. Algorithm of NACA Profile Design


approximation of source panel strength that cause
the body surface become a streamline of the flow.
The approximation can be made more accurate using
more panel numbers, and more closely representing
the source or vortex sheet of continuously varying
strength. The minimum number of panel that
produce accurate result can be obtained using panel
independence study.
The study conducted by simulating NACA 0012 at 5
degree angle of attack with various panel number,
from 10 to 160 panel numbers, with increment of 10.
While the parameter that is being monitored is CL
and CD.
II.5. NACA Profile Selection
To emphasize, the aim of the study is to get the most
efficient airfoil shape for two specific working
condition that is cruise and loiter. With wing area
and aircraft speed has been specified, the CL for
each condition has also been specified following
equation (5).
=

To avoid confusion with XFOIL terms, 2D


coefficient of lift and drag would be stated as CL and
CD instead of cl and cd. Unless being stated otherwise,
the term CL and CD in this study would be considered
for 2D.
II.6. Thrust Estimation
Recalling the equation (1b), the thrust required for
both cruise and loiter condition is equal to the drag
being generated. The wing platform and wing profile
selection would also yield drag value that can be
used for thrust estimation. The drag itself is obtained
using equation 6(a) (c).
= +
(6a)
1
2
= ( )
(6b)
2

= , ( 2 )
2

(6c)

III.
Result and Discussion
III.1. Wing platform, coefficient of lift and
Reynolds Number
Ten wing platform with different aspect ratio was
tested. 0.3 1.2 m wing span was chosen, giving a
range of aspect ratio from 0.7 to 11.1. For NACA
2412 and 2812, effect of aspect ratio to the total drag
is obtained and presented in Figure 2.
Figure 2 shows the total drag for both NACA 2412
and 2812 at various aspect ratio. Generally, it can be
observed that the total drag decrease as aspect ratio
increase for both cruise and loiter condition.
However, closer investigation reveals that a further
increase in aspect ratio above the value of 4
insignificantly decrease the total drag value. This
happened because the decrease in induced drag is
balanced by the increase in parasitic drag for a higher
aspect ratio. This insignificant decrease in total drag
is not beneficial compared to the structural trade-off
that is experienced by the longer-spanned wing.
Therefore, the aspect ratio below 4 is chosen. To be
specific a wing span of 0.6 m, producing aspect ratio
of 2.77 was chosen. A corresponding chord length
and Reynolds number for chosen wing platform is
shown in Table 2. The three-dimensional coefficient
of lift for both cruise and loiter condition, calculated
using equation (5), is also presented in Table 2.

(5)

While for certain airfoil shape, certain value of CL


can only be achieved by putting the airfoil at certain
angle of attack. Therefore the airfoil would be
mainly assessed based on the lowest drag at the
given CL. Also, pressure and friction distribution for
several airfoil shape is further examined in order to
get better understanding on the effect of thickness,
camber and camber position to aerodynamic
performance.
Being unrestricted in deciding NACA profile and
with so many options available, algorithm based on
good understanding in aerodynamics is important to
ensure the design process goes efficiently. The idea
of the algorithm is to decide an initial profile, and
then varying its thickness, maximum camber and
camber position consecutively, using the best
parameter from each stage to be used in the next one.
The decision in initial profile was based on the
understanding of aerodynamics, and would
determine the overall result. If other algorithm or
initial design is used, it should be noted that
optimum design obtained would be different. The
algorithm used in this study is shown in Figure 1.
3

Figure 2. Total drag for NACA 2412 and NACA


2812 airfoil at various wing aspect ratio
Table 2. Chosen wing platform and corresponding
Reynolds number and coefficient of lift

III.2. Panel independence study


Figure 3(a) shows value of cl and cd for both inviscid
and viscous condition. It can be observed that below
application of 60 panel numbers, the parameters
being monitored show different values for different
panel numbers. Low inviscid cl value of -0.04 was
recorded at the application of 10 panel numbers, and
rose up to 0.013 at 20 panel numbers. The fluctuation
continues until the application of 60 panel numbers,
where it reached a near-zero value. The similar trend
can be found for all other parameters. For the
application of 60 panel numbers or more, the
parameter values did not show any significant
difference. From this finding, it can be inferred that
60 is the minimum number of panels that enables the
simulation to give accurate result. The number of
panels strongly relates to the number of source or
vortex strength that can be approximated over the
airfoil surface. The larger the number, the finer the
distribution of those source/ vortex panel can be
obtained hence more accurate approximation. For
small number of panels, the distribution over airfoil
would only be captured partially. This might resulted
in a missing information or uncaptured phenomenon
within the result obtained. Prove of this occurrence
is shown in Figure 3(b) and (c).

Figure 3 (a) effects of panel number to cl and cd;


(b) Effect of number of panel on pressure
distribution; (c) skin friction for various number of
panel
Figure 3(b) shows the pressure distribution over
airfoil. The inviscid pressure distribution obtained
from simulation with 80 panels is used as the
reference. It can be seen that the viscous pressure
distribution for simulation using 20 panels
presented by dash-dot line shows negligible
separation phenomenon. Meanwhile for application
of 30 and 80 panels, separation observed at around
10% chord length from the leading edge, and
4

reattachment is observed at around 30% from the


leading edge. The phenomenon can be more clearly
observed by plotting the coefficient of friction over
the airfoil. As shown by Figure 3(c), application of
20 panel numbers shows no separation with positive
value of coefficient of friction observed along the
airfoil. Meanwhile at application of 30 and 80 panel
numbers, zero coefficient of friction detected from
10% to 30% chord length. As zero coefficient of
friction means that there is no flow that attached and
give friction to the airfoil surface, it can be used as a
good tool to predict flow separation and
reattachment.
Since viscous effects that corresponds to skin
friction and other boundary layer phenomenon are
ignored for the inviscid flow, the calculation for
forces around airfoil for this condition is simpler
compared to viscid one. This should lead to the need
of less panel to get accurate simulation output.
However, the result cannot clearly prove this
assumption, since both inviscid and viscous result
become independence to panel number at relatively
same panel number value.

Figure 4. Drag polar for 1% camber 20% camber


position airfoil with various thickness
For the same lift coefficient, the drag coefficient
produced was 0.007, 0.008, 0.0095 and 0.013 for
thickness of 6%, 9% 12% and 21% consecutively.
The relation between thickness and aerodynamic
performance can be further investigated by
observing Figure 5.
Figure 5 compares the pressure coefficient and skin
friction coefficient over airfoil with three different
thickness which are 3%, 12% and 21%. The airfoils
has same camber and camber position, and tested at
same angle of attack. The figure shows that the
thinnest NACA 1203 airfoil experience negligible
separation at the leading edge, and the flow stay
attached along the chord length. The flow is also
laminar at all airfoil surface, shown by the transition
point of 1 at the top and the bottom part of airfoil.
This produce excellent aerodynamic performance.
Meanwhile laminar separation bubble (LSB) was
observed at the upper part of airfoil for 12%thickness NACA 1212. It can be seen that the flow
started to separate from the leading edge, experience
transition at around 60% from the chord and then
reattach. The LSB is detected from the viscous
pressure coefficient moves away from the inviscid
pressure coefficient, and the near-zero skin friction.
This separation bubble decrease the aerodynamic
performance of the airfoil. Longer and deeper
separation bubble is observed at the 21%-thickness
NACA 1221 airfoil. It can also be observed that the
separation bubble formed at the upper as well as the
lower part of airfoil. This phenomenon decrease the
aerodynamic performance even further.
In such a low Reynolds number, which become the
case of the study, the separation bubble formed due
to the significant adverse pressure gradient. The
insufficient energy at the flow cannot overcome the
pressure gradient and then separates. At thicker
airfoil, higher adverse pressure gradient is apparent
because of the more curvature surface. Hence in the

III.3. NACA profile


Generally, low thickness airfoils is expected to be
thin since increased thickness would hampers the
aerodynamic characteristic, as observed by Okamoto
[5] and Brusov [1]. Therefore, initial thickness for
airfoil design would use a thin airfoil, which is 3%
chord length for cruise condition and 6% chord
length for loiter condition. Cruise condition has a
relatively high Reynolds number and low lift
coefficient compare to loiter condition, so it is
expected to be small cambered fly at small angle of
attack. Therefore, NACA 12xx was chosen to be the
initial design for cruise condition. Meanwhile, loiter
condition has lower Reynolds number and much
higher lift coefficient. Therefore, large cambered
airfoil is expected and NACA 42xx is chosen to be
the initial design.
i. Cruise condition
First in order to investigate the effect of thickness,
aerodynamic characteristics of 5 NACA profiles
namely 1203, 1206, 1209, 1212 and 1221 are
investigated and shown in Figure 4. Figure 4 shows
the drag polar curve that presents drag coefficient,
CD versus the lift coefficient, CL for the 5 NACA
profiles at angle of attack between -1 to 4. From the
graph, it can be seen that NACA 1203 gave lowest
drag coefficient of less than 0.007 for the target CL
of 0.31. It is also observed that the increase in
thickness decrease the aerodynamic performance.

Figure 5. Pressure and skin friction coefficient for NACA 1203, 1212 and 1221
cruise condition, thicker airfoil would likely to form
more separation bubble, which is undesirable.
Therefore, the 3% thickness is chosen for the cruise
condition.
The next step is to find the maximum camber. One
non-cambered and four different cambered airfoils
of 1%, 2%, 3% and 4% camber is tested. The
thickness used was the optimum thickness that is
obtained from previous investigation, which is 3%.
While the camber position is 20% of chord from the
leading edge. The airfoil was tested at a range of
angle of attack and the result is presented in Figure
6.

Figure 6 shows the drag polar of 5 airfoils at a range


of angle of attack. It is apparent that for the targeted
CL of 0.31, NACA 1203 produced lowest drag
coefficient of nearly 0.007. Meanwhile for higher
camber, higher value of drag was obtained to
produce same amount of targeted lift. The noncambered airfoil also give a high value of drag
coefficient for the targeted lift. To investigate the
effect of maximum camber, pressure coefficient
curve and skin friction curve for three different
airfoil is plotted and shown in Figure 7.
Figure 7 shows the pressure and skin friction
coefficient for NACA 0003, 2203 and 3203, all in
their own specific angle of attack that produce lift
coefficient of 0.31. To produce lift coefficient of 0.3,
NACA 0003 needed to set 2.76 o of angle of attack.
This value is a rather high angle of attack for such
thin non-cambered airfoil. Therefore, the separation
and transition happened very early at the leading
edge and showed stalling phenomenon. This
occurrence leads to a low lift to drag ratio.
Meanwhile the NACA 2203 produce lift coefficient
of 0.31 at the 1.2o AoA. Both pressure and skin
friction coefficient curve shows that neither
separation bubble nor transition occurred along the
airfoil. This is desirable and produce a good
aerodynamic performance. In the other hand, NACA
3203 produce lift coefficient of 0.31 at a small, 0.223
angle of attack. This small angle of attack at a

Figure 6. Drag polar for 3% airfoil thickness with


various camber

Figure 7. Pressure and skin friction coefficient for NACA 0003, 2203 and 3203
relatively high cambered airfoil induce the transition
to occur at the lower part of the airfoil. As can be
seen from the figure, the transition point for the
lower part of NACA 3203 airfoil is 0.0286 at the
leading edge. This resulting in poorer L/D value. By
revisiting Figure 6, it is clear that the cambered
airfoils are more suitable to obtain higher CL as it
gave lower drag at higher lift. While for lower CL,
small cambered airfoil is preferable.
Finally, after the optimum thickness and maximum
camber is obtained, optimum camber position was
investigated. Four camber position of 20%, 40%,
60% and 80% of chord length from the leading edge
was tested and the result is presented in Figure 8.

Figure 8 shows the drag polar for NACA 2203, 2403,


2603 and 2803. It is clearly observed that the further
the camber moves to the trailing edge, the higher the
drag is produced for the targeted lift. From the
pressure and skin friction coefficient graph shown in
Figure 9, it can be seen that the camber position
affects the pressure gradient along the airfoil. When
camber is closer to the leading edge for NACA 2203,
the high pressure gradient is observed at the front
part of the airfoil. The high pressure gradient then
moves backward as the camber position moves
further to the trailing edge. From the investigation, it
can be inferred that the 20% from the leading edge
is the optimum camber position.
The investigation in the effect of thickness, camber
and camber position shows that NACA profile of
1203 is the optimum profile for cruise condition.
ii. Loiter condition
The same approach is used to get the optimum
NACA profile for the loiter condition. To investigate
the effect of thickness, 4% cambered, 20% camber
position airfoil with different thickness are tested
and the result is shown in Figure 10.
Figure 10 shows the drag polar for NACA 4206,
4209, 4212 and 4221. It is vividly shown by the
figure that the increase in thickness deteriorate the
aerodynamic performance of airfoil. To get the same
targeted CL value, thicker airfoil produced higher

Figure 8. Drag polar of 2% camber 3% thickness


airfoil with various camber position
7

Figure 9. Pressure and skin friction coefficient for NACA 2203, 2603 and 2803
drag coefficient. The effect of thickness is further
investigated by looking at Figure 11, where the
pressure and skin friction coefficient of 6%, 12% and
21% thickness airfoil for the same angle of attack is
plotted. It is apparent that, while all the three airfoils
show the same tendency to produce laminar
separation bubble, thicker airfoil produce higher
skin friction especially near the leading edge.
Moreover, while the separation bubble is followed
by reattachment at NACA 4206, the separation
bubble is followed by a further turbulent separation
especially at NACA 4221. This leads to the lower
L/D value for thicker airfoil. Therefore, the 6% is
chosen to be the optimum thickness.

Meanwhile effect of camber is investigated by


plotting drag polar of four airfoils with different
camber, as shown in Figure 12. From the figure, it
can be inferred that NACA 4206 and 6206 produce
a similar, lowest drag coefficient. A closer
investigation shows that NACA 6206 produce
slightly lower L/D at the same CL and thus chosen to
be the optimum design. The effect of camber is
closer observed by Figure 13. To produce the same,
relatively high CL of 1.09, small cambered NACA
1206 needs AoA of 25.41o which produce stall and a
very low L/D value. While for the same C L, airfoil
profile with larger camber needs lower AoA. AoA of
6.57o and 4.88o is observed for NACA 4206 and
6206 consecutively. This lower AoA reduce the
adverse pressure gradient that has to be overcame by
the flow, enables the flow to reattach easier and
brought the airfoil away from the stall effect.
Therefore, 6% is chosen to be the optimum camber.
Investigation of the camber position effect to
aerodynamic performance at loiter condition is
shown in Figure 14 and 15. Figure 14 presents the
drag polar of four different airfoils, showing that the
40% chord length from the leading edge yield lowest
L/D among the airfoil being tested. Meanwhile
Figure 15 shows that moving camber position from
20% to 40% chord length move the separation
bubble and transition point away from the leading
edge. Meanwhile, by moving the camber further to

Figure 10. Drag polar for 4% camber 20% camber


position airfoil with various thickness
8

Figure 11, Pressure and skin friction coefficient for NACA 4206, 4212 and 4221

Figure 12. Drag polar for 20% camber position 6% thickness airfoil with various camber
III.4. Thrust
Using equation (3) to calculate induced drag
coefficient and XFOIL to obtained parasitic drag
coefficient, and using equation 6(a)-(c), the Drag for
cruise and loiter condition can be calculated. The
drag for cruise condition is 0.32 N while drag for
loiter condition is 2.79 N. And by recalling equation
1(b), the calculated drag can also be used to
represent the thrust.

80% of chord length form the leading edge,


transition point moves to the leading edge and severe
separation occurred over the airfoil. This resulted in
a poor L/D value.
Finally, the investigation in the effect of thickness,
camber and camber position shows that NACA
profile of 6406 is the optimum profile for loiter
condition.

Figure 13. Pressure and skin friction coefficient for NACA 1206, 4206 and 6206

Figure 14. Drag polar for 6% camber 6% thickness airfoil with various camber position
Table 3. Final wing design for loiter and cruise condition

10

Figure 15. Pressure and skin friction coefficient for NACA 6206, 6406 and 6806
longer and deeper it becomes, induced an
undesirable drag raise.

IV. Conclusion
The optimum wing design of cruise and loiter
condition for Micro-Air-Vehicle (MAV) is obtained.
By considering induced drag and Reynolds number
effect, as well as structural issue, aspect ratio of 2.77
was chosen. For the cruise condition, NACA profile
of 1203 was found to be the optimum design, giving
L/D ratio of 48.43 at angle of attack of 1.85 degree.
While NACA profile of 6406 is obtained as the
optimum design, giving L/D as high as 69.65 at
angle of attack of 3.88 degree. The final design
specification is shown in Table 3.
Generally, airfoil with low thickness is suitable for
low Reynolds number application. This is due to the
lower adverse pressure gradient produced by thinner
airfoil, minimizing the separation and the occurrence
of laminar separation bubble. This result agrees with
other research [1] [5]. Meanwhile, high cambered
airfoils become effective to obtain high value of lift
coefficient. Therefore, higher cambered airfoil is
suitable for loiter condition while lower one is
suitable for cruise condition. In the other hand,
camber position control the distribution of pressure
gradient over the airfoil.
Laminar separation bubble and transition point
become one of the main concern for airfoil flow over
low Reynolds number. Laminar separation bubble
was found in several airfoil design and is proven to
decrease the performance of airfoil. The bubble, the

Reference
[1]

[2]

[3]

[4]

[5]

[6]

[7]
[8]
[9]

11

V. Brusov and V. Petruchik, Design Approach for


Selection of Wing Airfoil with Regard to Micro-UAVs,
in International Micro Air Vehicles Conference, 2011.
A. Suhariyono, J. H. Kim, N. S. Goo, H. C. Park and K.
J. Yoon, Design of precision balance and aerodynamic
characteristic measurement system for micro aerial
vehicles, Aerospace Science and Technology, vol. 10,
pp. 92-99, 2006.
T. J. Mueller, Aerodynamic Measurements at Low
Reynolds Numbers for Fixed Wing Micro-Air
Vehicles, Defense Technical Information Center
Compilation Part Notice ADP010760.
N. Ismail, A. Zulkifli, M. Abdullah, M. H. Basri and N.
S. Abdullah, Optimization of aerodynamic efficiency
for twist morphing MAV wing, Chinese Journal of
Aeronautics, vol. 27, no. 3, pp. 475-487, 2014.
W. Shyy, Y. Lian, J. Tang, D. Viieru and H. Liu,
Aerodynamics of Low Reynolds Number Flyers,
Cambridge University Press, 2007.
U. Rist and K. Augustin, Control of Laminar
Separation Bubbles, Institut fur Aerodynamik and
Gasdynamik, Stuttgart.
M. S. Selig, Low Reynolds Number Airfoil Design,
VKI Lecture Series, 24-28 November 2003.
M. Drela and H. Youngren, XFOIL 6.94 User Guide,
MIT Aero & Astro, 2001.
J. D. Anderson, Fundamentals of Aerodynamics,
McGraw-Hill Education, 2010.

También podría gustarte